38
Implications of 21 st Century Climate Change for the Hydrology of Washington State Marketa M Elsner 1 , Lan Cuo 2 , Nathalie Voisin 2 , Jeffrey S Deems 2 , Alan F Hamlet 1,2 , Julie A Vano 2 , Kristian EB Mickelson 2 , Se-Yeun Lee 2 , and Dennis P Lettenmaier 1,2 Abstract T he hydrology of the Pacific Northwest (PNW) is particularly sensitive to changes in climate because seasonal runoff is dominated by snowmelt from cool season mountain snowpack, and temperature changes impact the balance of precipitation falling as rain and snow. Based on results from 39 global simulations performed for the Fourth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC AR4), PNW temperatures are projected to increase an average of approximately 0.3°C per decade over the 21 st century, while changes in annual mean precipitation are projected to be modest, with a projected increase of 1% by the 2020s and 2% by the 2040s. Based on IPCC AR4 projections, we updated previous studies of implications of climate change on the hydrology of the PNW. In particular, we used results from 20 global climate models (GCMs) and two emissions scenarios from the Special Report on Emissions Scenarios (SRES): A1B and B1. PNW 21 st century hydrology was simulated using the full suite of GCMs and 2 SRES emissions scenarios over Washington, as well as focus regions of the Columbia River basin, the Yakima River basin, and those Puget Sound river basins that supply much of the basin’s municipal water supply. Using two hydrological models, we evaluated projected changes in snow water equivalent, seasonal soil moisture and runoff for the entire state and case study watersheds for A1B and B1 SRES emissions scenarios for the 2020s, 2040s, and 2080s. We then evaluated future projected changes in seasonal streamflow in Washington. April 1 snow water equivalent (SWE) is projected to decrease by an average of approximately 27-29% across the State by the 2020s, 37-44% by the 2040s and 53-65% by the 2080s, based on the composite scenarios of B1 and A1B, respectively, which represent average effects of all climate models. In three relatively warm transient watersheds west of the Cascade crest, April 1 SWE is projected to almost completely disappear by the 2080s. By the 2080s, seasonal streamflow timing will shift significantly in both snowmelt dominant and transient, rain-snow mixed watersheds. Annual runoff across the State is projected to increase by 0-2% by the 2020s, 2-3% by the 2040s, and 4-6% by the 2080s; these changes are mainly driven by projected increases in winter precipitation. 1 JISAO Climate Impacts Group, University of Washington, Seattle, Washington 2 Department of Civil and Environmental Engineering, University of Washington, Seattle, Washington 3: Hydrology and Water Resources 69 CHAPTER 3: Hydrology and Water Resources: Washington State

3: Hydrology and Water Resources

  • Upload
    lenhan

  • View
    225

  • Download
    6

Embed Size (px)

Citation preview

Page 1: 3: Hydrology and Water Resources

Implications of 21st Century Climate Change for the Hydrology of Washington StateMarketa M Elsner1, Lan Cuo2, Nathalie Voisin2, Jeffrey S Deems2, Alan F Hamlet1,2, Julie A Vano2, Kristian EB Mickelson2, Se-Yeun Lee2, and Dennis P Lettenmaier1,2

Abstract

The hydrology of the Pacific Northwest (PNW) is particularly sensitive to changes in climate because seasonal runoff is dominated by snowmelt from cool season mountain snowpack, and temperature changes impact the balance of precipitation falling as rain and snow. Based on results from 39 global simulations performed for

the Fourth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC AR4), PNW temperatures are projected to increase an average of approximately 0.3°C per decade over the 21st century, while changes in annual mean precipitation are projected to be modest, with a projected increase of 1% by the 2020s and 2% by the 2040s. Based on IPCC AR4 projections, we updated previous studies of implications of climate change on the hydrology of the PNW. In particular, we used results from 20 global climate models (GCMs) and two emissions scenarios from the Special Report on Emissions Scenarios (SRES): A1B and B1. PNW 21st century hydrology was simulated using the full suite of GCMs and 2 SRES emissions scenarios over Washington, as well as focus regions of the Columbia River basin, the Yakima River basin, and those Puget Sound river basins that supply much of the basin’s municipal water supply. Using two hydrological models, we evaluated projected changes in snow water equivalent, seasonal soil moisture and runoff for the entire state and case study watersheds for A1B and B1 SRES emissions scenarios for the 2020s, 2040s, and 2080s. We then evaluated future projected changes in seasonal streamflow in Washington. April 1 snow water equivalent (SWE) is projected to decrease by an average of approximately 27-29% across the State by the 2020s, 37-44% by the 2040s and 53-65% by the 2080s, based on the composite scenarios of B1 and A1B, respectively, which represent average effects of all climate models. In three relatively warm transient watersheds west of the Cascade crest, April 1 SWE is projected to almost completely disappear by the 2080s. By the 2080s, seasonal streamflow timing will shift significantly in both snowmelt dominant and transient, rain-snow mixed watersheds. Annual runoff across the State is projected to increase by 0-2% by the 2020s, 2-3% by the 2040s, and 4-6% by the 2080s; these changes are mainly driven by projected increases in winter precipitation.

1JISAO Climate Impacts Group, University of Washington, Seattle, Washington2Department of Civil and Environmental Engineering, University of Washington, Seattle, Washington

3: Hydrology and Water Resources

69CHAPTER 3: Hydrology and Water Resources: Washington State

Page 2: 3: Hydrology and Water Resources

1. Introduction

The Fourth Assessment Report (AR4) of the Intergovernmental Panel on Climate Change (IPCC) states that warming of Earth’s climate is unequivocal and that anthropogenic use of fossil fuels has contributed to increasing carbon dioxide concentrations and thereby warming of the atmosphere (IPCC, 2007). The hydrology of the Pacific Northwest (PNW - which typically includes the Columbia River basin and watersheds draining to the Oregon and Washington coasts) is particularly sensitive to changes in climate because of the role of mountain snowpack on the region’s rivers. In this paper, we utilize archived climate projections from the IPCC AR4 to evaluate impacts on regional hydrology, with focus on Washington, which includes the lower Columbia River basin in the eastern and southern part of the State, as well as coastal drainages, including the Puget Sound basin (Figure 1).Washington is partitioned into two distinct climatic regimes by the Cascade Mountains. The west side of the Cascades on average receives approximately 1,250 mm of precipitation annually, while the east side receives slightly more than one-quarter of this amount. Washington, like much of the western US, relies on cool season precipitation (defined as October through March) and resulting snowpack to sustain warm season streamflows (defined as April through September). Approximately 75% of the annual precipitation in the Cascades falls during the cool season (Snover and Miles, in review). A changing climate affects the balance of precipitation falling as rain and snow and therefore the timing of streamflow over the course of the year. Figure 2 illustrates simulated historical mean annual runoff over the period 1916-2006 using the Variable Infiltration Capacity hydrologic model (further described below) and shows the importance of the State’s mountainous regions with respect to water supply for various natural resources.Small changes in temperature can strongly affect the balance of precipitation falling as rain and snow, depending on a watershed’s location, elevation, and aspect. Washington, and the Pacific Northwest as a whole, is often characterized as having three runoff regimes: snow-melt dominant, rain dominant, and transient (Hamlet and Lettenmaier 2007). In snowmelt dominant watersheds, much of the winter precipitation is stored in the snowpack, which melts in the spring and early summer resulting in low streamflow in the cool season and peak streamflow in late spring or early summer (May-July). Rain dominant watersheds are typically lower in elevation and mostly on the west side of the Cascades. They receive little snowfall. Streamflow in these watersheds peaks in the cool season, roughly in phase with peak precipitation (usually November through January). Transient watersheds are characterized as mixed rain-snow due to their mid-range elevation. These watersheds receive some snowfall, some of which melts in the cool season and some of which is stored over winter and melts as seasonal temperatures increase. Rivers draining these watersheds typically experience two streamflow peaks: one in winter coinciding with seasonal maximum precipitation, and another in late spring or early summer when water stored in snowpack melts. Hydrographs of simulated average historic streamflow, which are representative of the three watershed types, are shown in Figure 3.

70 CHAPTER 3: Hydrology and Water Resources: Washington State

Page 3: 3: Hydrology and Water Resources

Figure 1. Overview figure of Washington state, Puget Sound and Yakima case study basins, and significant analysis locations.

71CHAPTER 3: Hydrology and Water Resources: Washington State

Page 4: 3: Hydrology and Water Resources

Hydrologic simulations from which these hydrographs were developed are fully described in Section 2.2 below. The Chehalis River, which drains to the Washington coast (Figure 3), is a characteristic rain dominant watershed, while the Yakima River, which drains to the Columbia River (Figure 3), is a characteristic transient watershed, and the Columbia River, which drains from mountainous regions in mainly Canada, Idaho, Oregon, and Washington, is a characteristic snowmelt dominant watershed.Previous studies have presented metrics which can be used to define watershed type. Barnett et al. (2005) suggested a metric which they

Figure 2. Simulated mean annual runoff over Washington state by the Variable Infiltration Capacity (VIC) model over the historic period from 1916-2006.

Figure 3. Simulated monthly historic streamflow hydrographs for three representative watershed types in Washington, namely rain dominant (Chehalis River at Porter), transient rain-snow (Yakima River at Parker), and snowmelt dominant (Columbia River at the Dalles). Hydrographs represent monthly averages of simulated daily streamflow by the VIC model for 1916-2006.

72 CHAPTER 3: Hydrology and Water Resources: Washington State

Page 5: 3: Hydrology and Water Resources

defined as the ratio of peak snow water equivalent (SWE) to total cool season (October-March) precipitation. SWE is defined as the liquid water content of the snowpack. Barnett et al. (2008) also showed that SWE to precipitation ratios have been declining in the historic record due to observed warming, and that these changes are predominantly related to human influence on the climate. Hamlet and Lettenmaier (2007) characterized the three types of watersheds over the Pacific Northwest by temperature. Snowmelt dominant watersheds have average winter temperatures of less than -6°C., while completely rain dominant watersheds have average temperatures above 5°C. Their analysis explored changes in flood characteristics over basins of varying scale for these watershed categories. Hamlet (2007) and Mantua et al. (2009, this report) also applied the SWE to precipitation ratio metric to the Hydrologic Unit Code (HUC) 4 regions in the PNW as a means to catalogue high-disturbance areas. In Figure 4, we show the SWE to precipitation ratio computed for 1/16th degree grid cells over the PNW. Rain-dominant regions generally have ratios less than 0.1; transient regions are in the range of about 0.1-0.4; and, snowmelt dominant regions generally have ratios greater than 0.4 (see additional figures and discussion in Mantua et al., 2009, this report). Locations at which the historic streamflow hydrographs shown in Figure 3 were simulated are noted in Figure 4. Figure 4 shows that the urban water supply systems for the state’s major metropolitan areas in the Puget Sound basin and the Yakima area are located in transient regions. As shown in accompanying papers by Vano et al. (2009a; b, this report), shifts in seasonal streamflow in these regions toward higher winter flow and lower summer flow have strong implications for water management. This paper focuses on hydrologic impacts of climate change and relates those to the three watershed categories discussed above.

2. Approach and Methods

We applied a range of climate change projections from the IPCC AR4 (IPCC, 2007) to hydrologic model simulations and evaluated the impact of climate change on the hydrology of Washington with additional focus on the Columbia River basin, which is a major source of hydropower energy (Hamlet et al. 2009, this report), the Yakima River basin (Vano et al. 2009a, this report), which supports irrigation of high-valued crops such as orchards, and those Puget Sound watersheds that supply water to a majority of the state’s population (Vano et al. 2009b, this report). We performed hydrologic simulations using the Variable Infiltration Capacity (VIC) macroscale hydrology model (Liang et al., 1994; Nijssen et al., 1997) at 1/16th degree latitude by longitude spatial resolution over the entire state. We also applied the DHSVM, the Distributed Hydrology Soil and Vegetation Model (Wigmosta et al, 1994), at 150 meter spatial resolution over the Puget Sound watersheds. We used these models to explore sensitivity of runoff to changes in precipitation and temperature over our focus regions. We then evaluated implications of projected changes in snowpack and soil moisture over the same domains.

73CHAPTER 3: Hydrology and Water Resources: Washington State

Page 6: 3: Hydrology and Water Resources

Figure 4. The average ratio of peak VIC model simulated snow water equivalent (SWE) to October – March precipitation for the historical period (1916-2006).

74 CHAPTER 3: Hydrology and Water Resources: Washington State

Page 7: 3: Hydrology and Water Resources

2.1. Hydrologic Simulations

Studies of the impacts of climate change on regional hydrology are becoming increasingly common (Maurer, 2007; Maurer and Duffy, 2005; Hayhoe et al., 2007; Christensen and Lettenmaier, 2007; Christensen et al., 2004; Payne et al., 2004; Van Rheenen et al., 2004; Miller et al., 2003; among others). Many of these studies use a scenario approach which evaluates projections of hydrological variables, like streamflow, using a hydrology model forced with downscaled ensembles of projected climate from GCMs. These future climate simulations are then compared with a baseline hydrological simulation using historical climate (see e.g. Christensen and Lettenmaier, 2007; Maurer 2007; Hayhoe et al., 2007; among others). This approach is sometimes termed “off-line” forcing of a hydrological model, because it does not directly represent feedbacks between the land surface and climate system. An alternative approach, based on regional climate models, represents land-atmosphere feedbacks; however, complications arise due to bias in the climate model simulations (see Wood et al., 2004 for a detailed discussion), and computational requirements which generally preclude the use of multi-model ensemble methods. For this reason, we used the off-line simulation approach.We used climate change scenarios to force two hydrology models – the VIC Model (Liang et al., 1994; Liang et al., 1996; Nijssen et al., 1997) and DHSVM (Wigmosta et al. 1994). The VIC model is a macroscale model, meaning it is intended for application to relatively large distributed areas, typically ranging from 10,000 km2 or so, up to continental and even global scales. A key underlying model assumption is that sub-grid scale variability (in vegetation, topography, soil properties, etc.) can be parameterized, rather than represented explicitly. We evaluated VIC model simulations over all of Washington (and over the entire PNW to evaluate streamflow in the lower Columbia basin), including the Yakima River basin, which covers 15,850 km2.DHSVM is an explicitly distributed hydrology model, intended for application at much higher spatial resolution (and hence to smaller areas) than VIC. In this study, we applied DHSVM to relatively small rivers flowing to the Puget Sound basin at a 150 m spatial resolution. These watersheds range from 52 – 1055 km2 in area. Both VIC and DHSVM are described in more detail below.

2.1.1. Variable Infiltration Capacity (VIC) Model

The VIC model (Liang et al., 1994; Liang et al., 1996; Nijssen et al., 1997) has been used to assess the impact of climate change on U.S. hydrology in a number of previous studies. Hamlet and Lettenmaier (1999) studied the implications of GCM projections from the second IPCC assessment (1995) over the Columbia River Basin. Following the third IPCC Assessment Report (2001), Payne et al. (2004) studied climate change effects on the Columbia River Basin, Christensen et al. (2004) studied effects on the Colorado River, and Van Rheenen et al. (2004) studied effects on California. Similarly, recent studies by Vicuna et al. (2007) and Maurer (2007) analyzed the effects of IPCC AR4 projections on hydrologic systems in California, Christensen and Lettenmaier (2007)

75CHAPTER 3: Hydrology and Water Resources: Washington State

Page 8: 3: Hydrology and Water Resources

on the Colorado River basin, and Hayhoe et al. (2007) on the northeastern U.S., all using the VIC model.Although predictions of winter precipitation changes over the PNW have differed somewhat among recent IPCC reports (the 1995 report suggests an increase, whereas the 2001 report indicates only modest changes), warmer temperatures in all previous assessments have led to projections of reduced snowpack, and hence a transition from spring to winter runoff (Hamlet and Lettenmaier, 1999; Payne et al., 2004). Other impacts common to previous studies of hydrological impacts of climate change in the PNW include earlier spring peak flow and lower summer flows. In this paper, we used GCM simulations archived for the IPCC AR4 and increased the spatial resolution of the hydrological model over the PNW from 1/8th degree (used in all previous studies cited above) to 1/16th degree. An historical input data set including daily precipitation, maximum and minimum daily temperature, and windspeed was developed for this study at 1/16th degree spatial resolution and its unique features are described in section 2.2.1. Model calibration at routed streamflow locations included use of initial parameters for the 1/8th degree VIC model (Matheussen et al., 2000), transferred to the 1/16th degree model. These parameters were evaluated at 1/16th resolution at two calibration locations (Table 1a). Further calibration was performed over the Yakima River basin. Model calibration and validation statistics for the VIC model used in this study are provided in Table 1a and include relative error in mean annual streamflow and Nash Sutcliffe efficiencies. A well calibrated model typically yields a relative error less than 10% and a Nash Sutcliffe efficiency higher than 0.7 (Liang et al., 1996; Nijssen et al., 1997). Calibration and validation periods were chosen to include a range of streamflow conditions with which to test model performance. Other parameters (e.g. simulated SWE or soil moisture) were not used to further constrain model parameters. However, previous studies comparing VIC simulated SWE with observations (Andreadis et al., 2009 in review) and soil moisture with observations (Maurer et al., 2002), indicate that the model successfully simulates grid level processes. In addition to increasing the VIC model resolution for this study, the number of GCMs from which the ensembles are formed was increased substantially relative to previous studies.We also adapted the model to allow output of potential evapotranspiration (PET) for each model grid cell. PET is the amount of water that would be transpired by vegetation, provided unlimited water supply, and is often used as a reference value of land surface water stress in characterizations of climate interactions with forest processes (e.g., Littell et al., 2009, this report). PET is calculated in the VIC model using the Penman-Montieth approach (Liang et al., 1996) and the user may choose to output PET of natural vegetation, open water PET, as well as PET of certain reference agricultural crops.

2.1.2. Distributed Hydrology Soil Vegetation Model (DHSVM)

DHSVM was originally designed for application to mountainous forested watersheds, and includes explicit representations of the effects of forest vegetation on the water cycle, in particular the role of vegetation as it intercepts liquid and solid precipitation, and on snow accumulation and

76 CHAPTER 3: Hydrology and Water Resources: Washington State

Page 9: 3: Hydrology and Water Resources

ablation under forest canopies. Early applications of the model addressed how forest harvest affected flood frequency in the PNW (Bowling, 2000, La Marche and Lettenmaier, 2001; Bowling et al., 2001). The model represents runoff primarily via the saturation excess mechanism and explicitly represents the depth to water table at each model pixel, which has typically ranged from 10-200 m in past applications of the model (in our application to the Puget Sound basins, we used 150 m spatial resolution). Some DHSVM model parameterizations are similar to those in Topmodel (Beven and Kirkby, 1979); a key difference is the explicit, rather than statistical representation of downslope redistribution of moisture in the saturated zone. In addition to its representation of the water table and downslope redistribution of moisture, DHSVM represents the land surface energy balance (in a manner similar to VIC), unsaturated soil moisture movement, saturation overland flow, and snowmelt and accumulation. DHSVM simulates snow accumulation and ablation, using the same snow model used by VIC, which is described by Cherkauer et al. (2003). In brief, it uses a two-layer snow algorithm, in which the top layer is used to solve an energy balance with the atmosphere, including effects of vegetation cover, while the bottom layer is used as storage to simulate deeper snowpack. Using a 150 m resolution digital elevation model (DEM) as a base map (US Department of Interior/US Geological Survey, http://seamless.usgs.gov), DHSVM explicitly accounts for soil and vegetation types and stream channel network and morphology. Wigmosta et al. (1994; 2002) provide a detailed description of the model. The model also uses a soil class map based on the STATSGO soil map produced by the U.S. Department of Agriculture. The land cover map was derived from Alberti, et al. (2004). The model is forced by climate inputs including precipitation and temperature, (at daily or shorter time steps), downward solar and longwave radiation, surface humidity, and wind speed. Using the historical 16th degree dataset developed for the VIC model (described below) and procedures developed by Nijssen et al. (2001), daily forcings were disaggregated to 3-hour

Basins (gage)

Annual mean N-S model efficiency

Nat. (cms)

Sim. (cms)

Rel. error (%)

Calibration (monthly)

Validation (monthly)

Yakima (12505000) Calibration period (1986-2000) Validation period (1971-1985)

132.8 142.8 7 0.71 0.65

Columbia (14105700)Calibration period (1986-1999)Validation period (1970-1985)

5132 5375 4.5 0.85 0.83

Table 1a. Summary statistics of model calibration and validation for the Variable Infiltration Capacity (VIC) model in units of cubic meters per second (cms). The relative error is defined as the ratio of the difference between mean annual simulated flow (sim.) and mean annual observed natural flow (nat.) to the mean annual observed natural flow. The Nash Sutcliffe efficiency is a coefficient which is commonly used to define hydrologic predictive power, where a coefficient of one is a perfect match between simulated and observed natural flow.

77CHAPTER 3: Hydrology and Water Resources: Washington State

Page 10: 3: Hydrology and Water Resources

intervals as described in detail by Cuo et al. (2008b), who applied DHSVM to the entire Puget Sound basin. Model calibration and validation statistics for the DHSVM used in this study are provided in Table 1b. Similar to VIC, a well calibrated DHSVM model typically yields a relative error less than 10% and a Nash Sutcliffe efficiency higher than 0.7 (Wigmosta et al., 1994; Leung et al.,1996). Calibration and validation periods were chosen to include a range of streamflow conditions with which to test model performance.

2.2. Model Input Variables2.2.1. Historical Inputs

Both VIC and DHSVM require as forcing variables precipitation (Prcp) and temperature at a sub-daily time step, as well as downward solar and longwave radiation, surface wind, and vapor pressure deficit. All simulations described in this paper are based on a 1/16th degree spatial resolution data set of daily historical Prcp and daily temperature maxima and minima (Tmax, Tmin) developed from observations following methods described in Maurer et al. (2002) and Hamlet and Lettenmaier (2005), adapted as described below. Variables other than daily precipitation and temperature maxima and minima are derived from the daily temperature range or mean temperature following methods outlined in Maurer et al. (2002). One exception is surface wind. Daily wind speed values for 1949-2006 were downscaled from National Centers for Environmental Prediction-National Center for Atmospheric Research (NCEP-NCAR) reanalysis

Basins (gage)

Annual mean N-S model efficiency

Nat. (cms)

Sim. (cms)

Rel. error (%)

Calib. (daily)

Calib. (monthly)

Valid. (monthly)

Snohomish (12141300) Calibration period (1993-2002) Validation period (1983-1993)

35.5 36.1 2 0.50 0.79 0.75

Cedar (12115000) Calibration period (1982-1992) Validation period (1992-2002)

6.85 6.18 -10 0.61 0.81 0.81

Green (12104500) Calibration Period (1973-1983) Validation Period (1983-1993)

9.79 9.76 0 0.54 0.72 0.71

Tolt (12147600) Calibration period (1983-1993) Validation period (1993-2002)

1.52 1.39 -9 0.45 0.70 0.75

Table 1b. Summary statistics of model calibration and validation for the Distributed Hydrology Soil and Vegetation Model (DHSVM) in units of cubic meters per second (cms). The relative error is defined as the ratio of the difference between mean annual simulated flow (sim.) and mean annual observed natural flow (nat.) to the mean annual observed natural flow. The Nash Sutcliffe efficiency is a coefficient which is commonly used to define hydrologic predictive power, where a coefficient of one is a perfect match between simulated and observed natural flow.

78 CHAPTER 3: Hydrology and Water Resources: Washington State

Page 11: 3: Hydrology and Water Resources

products (Kalanay et al., 1996). For years prior to 1949, daily wind speed climatology was derived from the 1949-2006 reanalysis.We used the National Climatic Data Center Cooperative Observer (Co-Op) network and Environment Canada (EC) daily station data as the primary sources for precipitation and temperature values. We used a method described by Hamlet and Lettenmaier (2005) that corrects for temporal inhomogeneities in the raw gridded data using a set of temporally consistent and quality controlled index stations from the U.S. Historical Climatology Network (HCN) and the Adjusted Historical Canadian Climate Database (AHCCD) data. This approach assures that no spurious trends are introduced into the gridded historical data as a result of inclusion of stations with records shorter than the length of the gridded data set. The data are adjusted for orographic effects using the PRISM (Daly et al., 1994; 2002) climatology (1971-2001) following methods outlined in Maurer et al. (2002).Daily station data from 1915 to 2006 were processed as in Hamlet and Lettenmaier (2005), but using only Co-Op, EC, HCN, and AHCCD stations within a 100 km buffer of the domain. Quality control flags included in the raw Co-Op data set for each recorded value were used to ensure accuracy and to temporally redistribute “accumulated” Prcp values. We used the Symap algorithm (Sheppard, 1984; as per Maurer et al., 2002) to interpolate Co-Op/EC station data to a 1/16th degree.We then adjusted the daily Prcp, Tmax, and Tmin values for topographic influences by scaling the monthly means to match the monthly PRISM climate normals from 1970-2000. In the temperature rescaling method used for this study, Tmax and Tmin were adjusted by the same amount to avoid introducing a bias into daily mean temperatures and the daily temperature range. First, the average of the Tmax and Tmin values were computed for each of the monthly PRISM and monthly mean Co-Op time series. The difference between these values was applied as an offset to the average of the daily Tmax and Tmin in the appropriate month, thereby explicitly preserving the daily temperature range. For days where Tmin exceeds Tmax due to interpolation errors in the initial regridding step, we offset the average of these inverted Tmax and Tmin values and applied a climatological daily range from PRISM Tmax and Tmin.

2.2.2. Regional Climate Change Projections

As part of the IPCC AR4, results from a common set of simulations of 21st century climate were archived from 21 global climate models (GCM) (Mote and Salathé 2009, this report), using greenhouse gas emissions scenarios as summarized in the IPCC’s Special Report on Emissions Scenarios (SRES) (Nakicenovic, 2000). Simulations were archived predominantly for three SRES emissions scenarios (A1B, B1, and A2) for most of the 21 GCMs, with A2 following the highest trajectory for future CO2 emissions at the end of the 21st century. We focus on A1B and B1 emission scenarios because these were simulated by the most GCMs and our study focuses on mid-century change, at which point none of the scenarios is consistently the highest. Following Mote and Salathé (2009, this report), we used output from 20 of the GCMs for which monthly gridded precipitation, temperature, and other variables were archived for SRES emissions scenario A1B, and

79CHAPTER 3: Hydrology and Water Resources: Washington State

Page 12: 3: Hydrology and Water Resources

19 for which the same variables were archived for emissions scenario B1. Mote and Salathé (2009, this report) summarize the GCM predictions of 21st century precipitation and temperature over the Pacific Northwest, and evaluate performance of the GCMs in reconstructing 20th century climate. The spatial resolution of the 20 models varies, but is generally about three degrees latitude by longitude; therefore, we downscaled the climate model output to the spatial resolution of a regional hydrology model as described below.

2.2.3. Downscaling Procedures

In general, the GCM output is at too coarse a spatial resolution to be meaningful for hydrological studies. Therefore, we downscaled the GCM output to 1/16th degree spatial resolution and applied a delta method approach to develop climate change scenarios with which to evaluate impacts (see e.g. Hamlet and Lettenmaier, 1999; Snover et al., 2003). In the delta method, projected changes in precipitation and temperature, as determined by GCM simulations, are applied to the historical record at the resolution of hydrologic models. We used regional projected monthly changes derived from a total of 39 climate ensembles (described in Section 2.2.2). We performed hydrologic simulations using the historical record perturbed by these monthly changes and then evaluated impacts of climate change on a number of hydrologic variables.There are three previously established ways to develop climate change scenarios based on GCM output and may be used in off-line hydrologic simulations. As noted above, the delta method simply applies changes in temperature and precipitation from the GCM to historical inputs or inputs derived from historical data, which in turn are used to force the hydrological model in the same way that simulations using historical forcings are performed. This approach was used by Hamlet and Lettenmaier (1999). The second approach uses transient projections of future climate from GCMs statistically downscaled to the spatial resolution of the watershed model and from a monthly to daily time step. This approach was used by Christensen et al., 2004 and Christensen and Lettenmaier (2007) in the Colorado River basin, Van Rheenen et al. (2004), Maurer and Duffy (2005) in the Sacramento and San Joaquin basins of California, Payne et al. (2004) in the Columbia River basin, and Hayhoe et al. (2007) over the northeastern U.S. All of these studies followed the bias correction and statistical downscaling (BCSD) approach described by Salathé et al. 2007, Wood et al. (2004), and Wood et al. (2002). The third approach is to utilize regional climate model simulations constrained by GCMs to drive hydrologic models. Significant resources are required to implement this approach, which have limited its use.The advantage of the BCSD approach is that it makes direct use of transient climate change scenarios and, therefore, incorporates projected changes in climate variability. There are, however, some key assumptions in the spatial and temporal downscaling that can complicate interpretation of results at sub-monthly (e.g., daily or weekly) time steps. In addition, evaluation of transient scenarios is complicated by the stochastic element of the transient climate variability. Full analysis of this effect requires a large number of ensemble members; however, most GCMs archive only a single transient

80 CHAPTER 3: Hydrology and Water Resources: Washington State

Page 13: 3: Hydrology and Water Resources

run, and even for those that archive multiple ensembles, the number is generally quite small. The primary advantage of the delta method approach is that it provides realistic temporal sequencing associated with the historic record, while avoiding bias in the GCM simulations. Another advantage is that climate change impacts may be evaluated in the context of historical events. However, the primary disadvantage is that we do not incorporate projected changes in climate variability by the GCMs into the hydrologic simulations. The delta method approach is arguably more appropriate for this study to evaluate water resource system performance at a sub-monthly timestep in a changing climate, as reported in companion papers by Hamlet et al. 2009 (this report) and Vano et al. 2008a; b (this report).We performed hydrologic simulations to evaluate the impacts of climate change on statewide hydrology in the 2020s, 2040s and 2080s. The delta values represent monthly average changes for each future period over the whole PNW and were applied to Washington. The PNW is arguably the smallest area that the GCMs are able to resolve and, therefore, potential differences in rates of climate change across the State are not incorporated. Each future period represents a 30-year average of projected climate, for instance, the 2020s are represented by the 30-year average climate between 2010 and 2039. Likewise for the 2040s and 2080s, these represent the average climate over 30-year periods 2030-2059 and 2070-2099, respectively. Six composite scenarios were formed following methods outlined by Mote and Salathé, 2009 (this report). In particular, for each 30-year time period and each month, we computed domain-average precipitation and temperature changes. Unlike Mote and Salathe (2009, this issue), we assume equal weighting of each climate change scenario for this study because, as similarly found by Brekke et al. (2004), the weighting of scenarios is largely dependent on the criteria used. In accordance with the delta method approach, we perturbed the entire spatially gridded record of observed historical daily precipitation and temperature (1916-2006) by the projected change for the corresponding month (12 values for each of precipitation and temperature), for each of the three future periods.In addition to performing hydrology simulations over the Washington using composite scenarios, we also performed simulations using 39 individual scenarios of 2020s climate over focus watersheds of the Yakima River basin and the Puget Sound for each of the GCMs. These focus watersheds are further described below.

2.3. Focus Watersheds

We evaluated in more detail the impacts of projected future climate change on the hydrology of three key areas: The Puget Sound drainage basin, the Yakima River basin, and the Columbia River basin. The three focus regions are shown in Figure 1.The Columbia River basin is one of our focus basins because it drains the eastern 2/3 of the state, as well as much of Idaho, part of British Columbia, and 2/3 of Oregon. In addition, roughly 70 percent of the electrical energy consumption within the State of Washington is derived from hydropower, most of which comes from the Columbia River (Bonneville Power Administration 1994). Detailed analysis of impacts on the Columbia River in the context of hydropower production are presented in a companion

81CHAPTER 3: Hydrology and Water Resources: Washington State

Page 14: 3: Hydrology and Water Resources

paper, Hamlet et al. (2009, this report).The Puget Sound basin is bounded to the east by the Cascades and to the west by the Olympic Mountains, and covers an area of approximately 30,000 km2. Its elevation ranges from sea level to 4,400 m. Substantial winter snowfall occurs at high elevations, but rarely in the lowlands. Annual precipitation ranges from 600 to over 3,000 mm, depending on elevation, most of which falls from October to March. The watersheds that make up the Puget Sound basin are generally characterized as transient. The Puget Sound basin includes more than 69% of the State’s population (based on 2000 census). Quantification of the region’s future water supply is therefore critical to the region’s future growth and ecosystem conservation. We focus here on four Puget Sound watersheds that are managed primarily for water supply: the Cedar River basin, Green River basin, Tolt River basin, and Sultan River basin (Figure 1). In a companion paper (Vano et al., 2008a, this report), we use the hydrological sequences described herein as input to reservoir simulation models. In this paper, we limit our attention to the hydrological projections.The Yakima River, which drains east through an arid lowland area, supplies water to over 180,000 irrigated hectares (450,000 acres). Agriculture in the Yakima River basin has changed over time. Land used to grow annual crops (e.g. wheat) has decreased, while that used to grow perennial crops including apples and grapes has increased. This shift toward perennial crops has increased dependence by agricultural producers on reliable water supplies (EES, Inc. 2003). Vano et al. (2008b) use the hydrological sequences described herein in conjunction with a reservoir simulation model of the Yakima River basin to evaluate potential climate change impacts on agricultural production in the basin.

3. Model Sensitivities to Changes in Climate

By the 2040s, future regional temperatures are projected to be out of the range of historic variability (Mote at Salathé 2009, this report). Further, we lack observations to evaluate the sensitivity of hydrologic models to projected changes in climate, which makes evaluating confidence in predicting impacts difficult. The need for “validation” of hydrological models is widely accepted in the hydrological literature, and it is usually performed by using split sample methods first to estimate model parameters, and then to evaluate model performance (see e.g. Refsgaard and Storm, 1996). However, a similar structure for evaluation of model sensitivities, such as how much runoff will change for a given amount of warming, is often lacking. Dooge (1992) suggested a framework for assessing hydrological sensitivity to climate change, via what is referred to as elasticity, or the fractional change in runoff compared to the fractional change in precipitation (precipitation elasticity) or potential evapotranspiration (PET elasticity). Here we focus on precipitation elasticity, which can be evaluated, on an annual basis, from historical observations of streamflow (or runoff) and precipitation. Simulated runoff may be used as a surrogate for streamflow in calculation of elasticity because, on an annual basis, the difference introduced by the time lag of streamflow routing is negligible. Previous studies show that precipitation elasticities performed on the same watershed using different hydrologic models can lead to different results. For example, the results

82 CHAPTER 3: Hydrology and Water Resources: Washington State

Page 15: 3: Hydrology and Water Resources

from Nash and Gleick (1991) and Schaake (1990) for the Colorado River differ in their precipitation elasticities by a factor of about two.Sankarasubramanian et al. (2001) suggested a non-parametric, robust, and unbiased elasticity estimator which summarizes sensitivity of streamflow to changes in precipitation, which yields similar results for a wide range of hydrologic model structures. Their estimator of the streamflow elasticity to precipitation is:

−−

=QP

PPQQmediane

t

tp

(1)

where Qt and Pt are annual streamflow and precipitation, respectively, and and are the long-term mean annual streamflow and precipitation.A result of the Sankarasubramanian et al. (2001) work was a contour map for the continental U.S. of (annual) streamflow elasticities to precipitation. The map shows streamflow elasticities in the range 1.0-2.0 for much of Washington State. In other words, a given fractional change in precipitation would result in a one- to two-fold fractional increase in streamflow. Using eq. 1, we evaluated observed and simulated runoff elasticities to precipitation for 6 locations within the Yakima watershed and 6 in the Puget Sound basin. These locations are noted in Figure 1 (overview map) and are defined in Table 1. Elasticities for the Yakima River watersheds were calculated using results from the VIC model, while elasticities for the Puget Sound were calculated using the DHSVM model.Analysis of temperature sensitivities is slightly more complicated. In the Dooge (1992) formulation, streamflow elasticities to precipitation and potential evaporation are described and these are used as a measure of model sensitivity. However, potential evaporation is a computed, rather than observed, quantity. In general, it depends on net radiation, vapor pressure deficit, wind, and land surface properties such as roughness length. Several of these quantities are temperature dependent. Furthermore, hydrological sensitivities to temperature are generally much more subtle than to precipitation, and they are difficult to estimate from observations because precipitation effects dominate the results. Instead, we computed runoff sensitivity to temperature in two ways. The first is a fixed temperature increase, in which both daily maximum and minimum temperature were increased by 1°C. In the Maurer et al. (2002) formulation of land surface forcing variables, downward solar radiation is indexed to the daily temperature range, hence for the same increase in Tmin and Tmax, downward solar radiation is constant (however, net longwave radiation, as well as vapor pressure deficit, both change). Such a fixed temperature increase was used to develop delta method scenarios in this study. The second computation also changes the daily average temperature by 1°C, but leaves Tmin unchanged, while increasing Tmax by 2°C. This has the effect of increasing downward solar radiation, but leaving the dew point (which is directly related to the daily minimum temperature in the model) unchanged. Meehl et al. (2007) summarizes projected changes in the global diurnal temperature range (i.e. difference between Tmax and Tmin). Although this range is expected to change over parts of the globe, there is no consensus among GCMs over the direction of change for the

83CHAPTER 3: Hydrology and Water Resources: Washington State

Page 16: 3: Hydrology and Water Resources

Pacific Northwest. Therefore, we decided that it is most appropriate to apply the delta method approach using fixed change in Tmax and Tmin.We analyzed precipitation elasticity and temperature sensitivities for six locations in the Yakima River basin, which correspond to the five basin reservoir locations, in addition to the Yakima River at Parker (USGS ID 12505000), which is a key reference station for water management in the basin. Observed and simulated precipitation elasticities calculated from the historical record for these sites are in close agreement and are summarized in Table 3. They range from 1.08 to 1.42 in the Yakima watershed. A 10% increase in precipitation causes an increase in runoff of a factor of 1.59 for the entire basin (at Parker) to 1.87 for Bumping Lake, which has a small contributing area (184 km2) and is at a relatively high elevation (1,030 m). An average daily temperature increase of 1°C, applied by increasing both minimum and maximum temperature (downward solar radiation unchanged), reduces basin runoff by approximately 2.45 (Rimrock) to 5.77% (Bumping Lake) (refer to Table 2, Temperature Sensitivity a). Alternatively, the same average daily increase, by altering maximum temperature only (constant dew point), reduces runoff by 5.15% (Parker) to 9.81% (Bumping Lake) (refer to Table 2, Temperature Sensitivity b).In the Puget Sound basin, we analyzed six catchments including the Cedar River at Renton, (Cedar E), the Cedar River near Cedar Falls (Cedar A), Green River near Auburn (Green C), the Green River above Howard Hanson Dam (Green A), the Sultan River (Sultan A) and the South Fork Tolt River near Index (Tolt A) (see Figure 1 for locations). These points are generally located near water supply reservoirs. Precipitation elasticity of observed and simulated historical periods at the six sites are in agreement (See Table 2) with values ranging from 1.0 – 1.4. An increase in precipitation of 10% for the same simulated watersheds (with temperature remaining unchanged) causes an increase in runoff by a factor of 1.17 to 1.63 in the Puget Sound basin. An average temperature increase of 1°C, by increasing both maximum and minimum temperature by 1°C (see Table 3, Temperature Sensitivity a), results in approximately a 0.7-2.4% decrease in the Puget Sound basin streamflows. The same average increase in daily temperature applied by increasing the maximum temperature by 2°C and leaving the minimum temperature unchanged (see Table 3, Temperature Sensitivity b) results in decreases in runoff by 1.5-5.6%.Runoff sensitivity to temperature change is expected to be higher when only Tmax is increased as compared with increasing both the Tmax and Tmin. This is expected because the algorithm used to estimate downward solar radiation is based on the daily temperature range, and therefore downward solar radiation, remains constant when both the maximum and minimum temperature are increased. As a result, the change in net radiation is generally smaller than when the minimum temperature is left unchanged. The basis for different precipitation elasticities and temperature sensitivities across sites is less clear. Elasticities are generally higher for Yakima River basin sites than for Puget Sound sites, but it is not entirely clear whether these differences are related to watershed characteristics or to potentially different sensitivities of the two hydrologic models. Precipitation and temperature sensitivities calculated above are based on annual changes and runoff responses will vary depending on the seasonality of change.

84 CHAPTER 3: Hydrology and Water Resources: Washington State

Page 17: 3: Hydrology and Water Resources

Site ID Description Basin Area (km2) USGS ID

Yakima Basin

BUMPI Bumping River near Nile 184 12488000

RIMRO Tieton River at Tieton Dam near Naches 484 12491500

KACHE Kachess River near Easton 166 12476000

KEEMA Yakima River near Martin 142 12474500

CLERO Cle Elum River near Roslyn 526 12479000

YAPAR Yakima River near Parker 6,889 12505000

Columbia Basin

DALLE Columbia River at the Dalles 613,827 14105700

Puget Sound Basin

Cedar E Cedar River at Renton 469 12119000

Green C Green River Outlet near Auburn 1032 NA

Cedar A Cedar River near Cedar Falls 106 12115000

Green A Green River above Howard Hanson Dam 573 NA

Sultan A Sultan River 178 NA

Tolt A South Fork Tolt River near Index 17 12147600

Table 2. Summary of analysis locations. Sites with USGS ID of “NA” indicate these are not USGS gage locations.

Site

Observed (and Simulated)

Precipitation Elasticity

Precipitation Elasticity

(10% increase)

Temperature Sensitivity (a),

%/oC

Temperature Sensitivity (b),

%/oC

Yakima Basin

BUMPI 1.42 (1.12) 1.87 -5.77 -9.81

RIMRO 1.37 (1.08) 1.65 -2.45 -6.26

KACHE 1.16 (1.23) 1.67 -3.70 -6.36

KEEMA 1.15 (1.19) 1.78 -5.19 -7.56

CLERO 1.12 (1.13) 1.61 -4.01 -6.73

YAPAR 1.32 (1.32) 1.59 -2.84 -5.15

Puget Sound Basin

Cedar E 1.38 (1.22) 1.36 -1.11 -2.99

Green C 1.33 (1.43) 1.63 -2.33 -5.57

Cedar A 1.08 (1.17) 1.28 -1.05 -2.77

Green A 1.42 (1.37) 1.61 -2.42 -5.64

Sultan A 1.06 (1.12) 1.17 -0.69 -1.69

Tolt A 1.12 (1.00) 1.20 -0.66 -1.50

Table 3. Summary of precipitation elasticity and temperature sensitivity at analysis locations. Precipitation elasticity is defined as the ratio of the fractional change in runoff to the fractional change in precipitation. Temperature sensitivities are defined as the percent change in runoff per 1°C of warming. Temperature sensitivity (a) considers increased daily minimum and maximum temperature, while temperature sensitivity (b) considers increased daily maximum temperature.

85CHAPTER 3: Hydrology and Water Resources: Washington State

Page 18: 3: Hydrology and Water Resources

4. Results and Discussion

Projections of 21st century climate of the PNW summarized in Mote and Salathé (2009, this report) indicate that temperatures will increase an average of 0.3°C (0.5°F) per decade. Changes in annual mean precipitation are projected to be modest, with a projected increase of 1% by the 2020s and 2% by the 2040s. However, the range of projected precipitation shows a decrease of almost 11% to an increase of almost 20% by the 2080s, underscoring the uncertainty in projections of future precipitation. Projected temperature increases, along with changes in seasonal precipitation have important implications for hydrologic variables across Washington. In this section we summarize impacts of projected changes in climate on a state level, as well as the Columbia River watershed, and then provide a more focused evaluation of watersheds within the Puget Sound and Yakima drainage basins.

4.1. Statewide Climate Change Impacts4.1.1. Implications of Changes in April 1 Snow Water Equivalent

Many past studies demonstrated that changes in snowpack are a primary impact pathway associated with regional warming in the PNW (Lettenmaier et al., 1999; Hamlet and Lettenmaier, 1999; Snover et al., 2003). Changes in snowpack are affected by both precipitation and temperature, although in the 20th century, temperature has been the more important driver (Mote et al., 2005; Hamlet et al., 2005; Mote, 2006; Mote and Salathé, 2009, this report), particularly in relatively warm areas such as the Cascades. SWE on April 1 is an important metric for evaluating snowpack changes because in the PNW, the water stored in the snowpack on April 1 is strongly correlated with summer water supply.Figure 5 shows projected changes in April 1 SWE for the 2020s, 2040s, and 2080s for the composite A1B and B1 climate conditions, as simulated using the VIC model. Results from these hydrologic simulations are consistent with previous studies, such as the climate impacts study conducted for King County, Washington, which projected a decrease in snowpack over the 21st century (Casola et al. 2005). Generally, results using the B1 emissions scenario project less significant impacts than those using the A1B scenario. Based on composite scenarios for the B1 and A1B scenarios respectively, April 1 SWE is projected to decrease by 27 to 29% across the state by the 2020s, 37 to 44% by the 2040s and 53 to 65% by the 2080s.Changes in SWE vary by elevation, as Figure 5 suggests. We summarized these changes over three bands of elevation, specifically elevations below 1,000 meters, between 1,000 and 2,000 meters, and above 2,000 meters (see Table 4). The results show that the lowest elevations will experience the largest decreases in snowpack, with reductions for B1 and A1B emissions scenarios, respectively, of 36 to 37% by the 2020s to 62 to 71% by the 2080s. The reduction of snowpack in the regions of highest elevation is projected to be less significant.Projected changes in snowpack are directly correlated with temperature. The greatest sensitivity of snowpack to warming is at temperatures near

86 CHAPTER 3: Hydrology and Water Resources: Washington State

Page 19: 3: Hydrology and Water Resources

Figure 5. Summary of projected April 1 snow water equivalent (SWE) for the 2020s, 2040s and 2080s (A1B and B1 SRES scenarios) by the VIC model. Percent change values represent spatially averaged April 1 SWE across Washington State.

87CHAPTER 3: Hydrology and Water Resources: Washington State

Page 20: 3: Hydrology and Water Resources

freezing. Locations with a warmer mean historical winter temperature (defined as December through February) are projected to experience the greatest reduction of snowpack, while locations with cooler winter temperatures are projected to experience more modest reductions (Figure 6). Projections using the A1B emissions scenario generally show greater reductions in snowpack than those using the B1 scenario, especially for the 2080s simulations

4.1.2. Implications of Changes in July 1 Soil Moisture

Vegetation and dry land agriculture rely heavily on soil moisture, in addition to precipitation, particularly in the arid region of the state (east of the Cascades in the Columbia River basin) where summer precipitation is low. Soil moisture in snow dominated watersheds (like the Columbia River basin overall) tends to peak in spring or early summer, in response to melting mountain snowpack. In the summer, lower precipitation (along with clearer and longer days) and increased vegetative activity cause depletion of soil moisture, resulting in minimum soil moisture values in September.Simulated soil moisture by hydrologic models is strongly determined by

Figure 6. Projected change in April 1 snow water equivalent (SWE) for the 2020s, 2040s, and 2080s plotted against mean historical winter temperature (1916-2006). Individual points represent April 1 SWE for each of the 91 years simulated by the VIC model. Projected values are derived using a delta method approach, where historical temperature and precipitation are perturbed by the projected average monthly changes in these vales for the 2020s (average change from 2010-2039), 2040s (average change from 2030-2059), and 2080s (average change from 2070-2099).

2020s 2040s 2080s

% Change in (2010-2039) (2030-2059) (2070-2099)

April 1 SWE A1B B1 A1B B1 A1B B1

< 1,000m (< 3,280ft) -37% -36% -54% -46% -71% -62%

1,000 m - 1,999 m(3,280 ft – 6,558 ft) -27% -25% -42% -34% -63% -51%

>= 2,000 m(>= 6,558 ft) -17% -15% -29% -23% -54% -39%

Overall -29 -27% -44% -37% -65% -53%

Table 4. Projected changes (%) in April 1 snow water equivalent (SWE) according to elevation using delta method composite climate change scenarios (30-year average changes not weighted) for the 2020s, 2040s, and 2080s.

88 CHAPTER 3: Hydrology and Water Resources: Washington State

Page 21: 3: Hydrology and Water Resources

model assumptions (Liang et al., 1998), but when expressed as percentiles, many of these differences are removed (Wang, 2008). For this reason, we present projected soil moisture changes across the state as percentiles of simulated historic mean soil moisture (1916-2006), where a projected decrease in soil moisture is represented by percentiles less than 50 and a projected increase is represented by percentiles greater than 50. Specifically, we summarize projections of July 1 soil moisture from the VIC model, as this is the typical period of peak soil moisture which is critical for water supply in the State’s arid regions.Projections of July 1 total soil moisture change for the composite A1B and B1 scenarios are modest but generally show decreases across the State. Projected decreases are greater for A1B scenario simulations compared with B1 simulations. For the three future periods, soil moisture is projected to be in the 38th to 43rd percentile (A1B and B1, respectively) by the 2020s, 35th to 40th percentile by the 2040s, and 32nd to 35th percentile by the 2080s, with 50% being equal to mean historical values. However, projected soil moisture changes vary on either side of the Cascade Mountains. In the mountains and coastal drainages west of the Cascades, warming of the climate tends to enhance soil drying in the summer and, in combination with reduced winter snowpack and earlier snowmelt, causes decreases in summer soil moisture (Figure 7). East of the Cascades, summer soil moisture is primarily driven by recharge of snowmelt water into the deep soil layers. Increased snowpack at the highest elevations in some parts of the Cascades (tied to projected increases in winter precipitation) and subsequently increased snowmelt, are likely to cause greater overall infiltration. Similar trends east and west of the Cascades were found in the study of PNW regional climate change impacts (Casola et al., 2005).

4.1.3. Implications of Changes in Mean Annual Runoff and Streamflow

As noted by Mote and Salathé (2009, this report), there is a wide range in projections of future precipitation across GCMs and SRES emissions scenarios. Across the 39 scenarios considered in this study (20 GCMs and 2 SRES emissions scenarios for all but one GCM), projected annual precipitation changes over the PNW range from -9% to +12% for the 2020s, -11% to +12% for the 2040s, and -10% to +20% for the 2080s, with modest increases projected for the composite scenarios for A1B and B1 (Table 5). Although projected increases of annual precipitation are modest, projections of seasonal precipitation change indicate increased winter precipitation and decreased summer precipitation (Tables 6 and 7). With 75 % of the annual precipitation falling between October and March (Snover and Miles, in review), cool season precipitation is the primary driver of hydrologic processes in Washington and the PNW. Projections of cool season precipitation for the composite B1 and A1B scenarios, respectively, range from +2.3% to +3.3% for the 2020s, +3.9% to 5.4% for the 2040s, and +6.4% to +9.6% for the 2080s (Table 6). Table 5 summarizes the composite projected changes in annual precipitation and corresponding state-wide changes in runoff simulated by the VIC model. The importance of cool season precipitation to the state’s runoff is evident: even with increased temperatures and modest, as opposed to significant, annual precipitation increases (and in the case of the 2020s for emissions scenario A1B, a slight decrease in annual precipitation) runoff

89CHAPTER 3: Hydrology and Water Resources: Washington State

Page 22: 3: Hydrology and Water Resources

Figure 7. Summary of projected July 1 soil moisture for the 2020s, 2040s, 2080s (A1B and B1 SRES scenarios) as percentile of simulated historical mean from 1916-2006 (using the VIC model). Percentiles less than 50 represent a decrease in soil moisture, while percentiles greater than 50 show an increase in soil moisture. Reported values represent spatially averaged percentile across Washington State.

90 CHAPTER 3: Hydrology and Water Resources: Washington State

Page 23: 3: Hydrology and Water Resources

2020s 2040s 2080s

(2010-2039) (2030-2059) (2070-2099)

A1B B1 A1B B1 A1B B1

Change in Temperature +1.1°C +1.0°C +1.8°C +1.4°C +3.2°C +2.3°C

% Change in Precipitation +2.3% +3.3% +5.4% +3.9% +9.6% +6.4%

% Change in Runoff +10.9% +12.6% +20.5% +16.1% +34.6% +25.6%

Table 6. Summary of composite changes in cool season (October through March) precipitation and runoff across Washington using delta method composite climate change scenarios (30-year average changes not weighted) for the 2020s, 2040s, and 2080s for SRES A1B and B1 global emissions scenarios.

2020s 2040s 2080s

(2010-2039) (2030-2059) (2070-2099)

A1B B1 A1B B1 A1B B1

Change in Temperature +1.3°C +1.2°C +2.3°C +1.7°C +3.8°C +2.7°C

% Change in Precipitation -4.2% -0.9% -5.0% -1.3% -4.7% -2.2%

% Change in Runoff -19.1% -15.8% -28.6% -22.1% -43.2% -33.4%

Table 7. Summary of composite changes in warm season (April through September) precipitation and runoff across Washington using delta method composite climate change scenarios (30-year average changes not weighted) for the 2020s, 2040s, and 2080s for SRES A1B and B1 global emissions scenarios.

2020s 2040s 2080s

(2010-2039) (2030-2059) (2070-2099)

A1B B1 A1B B1 A1B B1

Change in Temperature +1.2°C +1.1°C +2.1°C +1.6°C +3.5°C +2.5°C

% Change in Precipitation +0.2% +1.9% +2.1% +2.2% +4.9% +3.4%

% Change in Runoff 0.0% +2.3% +2.7% +2.2% +6.4% +4.2%

Table 5. Summary of composite changes in annual precipitation and runoff across Washington using delta method composite climate change scenarios (30-year average changes not weighted) for the 2020s, 2040s, and 2080s for SRES A1B and B1 global emissions scenarios.

91CHAPTER 3: Hydrology and Water Resources: Washington State

Page 24: 3: Hydrology and Water Resources

increases in all cases. This contrasts with results for precipitation elasticity and temperature sensitivities (Table 3) to the extent that on an annual basis, the modest precipitation changes coupled with temperature increases should have led to runoff reductions. The reason this is not the case is that in the Table 3 experiments, precipitation changes are uniform over the year, whereas in the GCM output (at least for the composites), cool season precipitation, which is much more efficient than summer precipitation in terms of runoff production due to higher soil moisture storage and lower vegetative water demand, increases while summer precipitation decreases.These results differ from the projected changes in runoff presented by Milly et al. (2005), who summarized average changes in runoff over Water Resources Regions across the continental U.S. and Alaska, defined by the U.S. Water Resources Council for the period 2041-2060, relative to 1901-1970. Their projections are based on output from 12 IPCC AR4 GCMs and the A1B SRES scenario, and showed slight decreases in runoff of 2-5% across the PNW. The 12 GCMs they used are a subset of the 21 (IPCC AR4) models used in this study. Milly et al. (2005) average over 24 ensembles from the 12 models (i.e. for some GCMs, multiple experiments were conducted on the same model); however, the number of ensembles was not the same for each GCM, which effectively weights some models more heavily than others. In addition, Milly et al. (2005) used land surface schemes embedded in the GCMs, which are at coarser resolution than the VIC model and do not resolve the topography of the PNW.Projections of streamflow differ from those of runoff because runoff is a spatial quantity that is an integral part of the water balance at each hydrologic model grid cell and does not incorporate the time lag effects that contribute to streamflow. Runoff is useful for evaluating projected basin-wide changes as a direct effect of precipitation and snow storage or melt. Streamflow, however, is the culmination of hydrologic processes evaluated at a given location over time. Figure 8 shows projected mean hydrographs for the example rain-dominant, transient, and snow-dominant watersheds in Figure 4. In the Chehalis River, projected changes to the mean hydrograph are minimal. Changes in the mean hydrograph at The Dalles are more apparent, including reduced peak flow in the late spring and early summer and increased cool season flow in connection with reduced snowpack. Changes in the Yakima watershed, a transient rain-snow watershed, are significant, indicating a shift to a characteristic rain-dominant watershed by the 2080s. Vano et al. (2009b, this report) describes the implications of this change on water management in the basin.

Figure 8. Projected average monthly stream flow for a rain dominant watershed (Chehalis River at Porter), transient rain-snow watershed (Yakima River at Parker), and snowmelt dominant watershed (Columbia River at The Dalles). Hydrographs represent monthly averages of simulated daily streamflow by the VIC model for the historic period (1916-2006) and three future periods (2020s, 2040s, and 2080s) using the A1B SRES scenario.

92 CHAPTER 3: Hydrology and Water Resources: Washington State

Page 25: 3: Hydrology and Water Resources

4.2. Hydrologic Case Studies

We evaluated impacts of climate change on three focus regions, namely the Columbia River Basin, the Puget Sound, and the Yakima River basin. Because the Columbia River basin covers approximately 2/3 of Washington State, discussion of impacts in this region is incorporated into the discussion of statewide impacts above. The other two case study watersheds, the Puget Sound and Yakima River basin, are discussed here. They are both transient watersheds, meaning they are highly sensitive to climate change; however, they differ with respect to their climatic regime – precipitation is generally much higher in the Puget Sound basin than in the Yakima, particularly its lower reaches. As noted in Section 2.2, we used the high resolution DHSVM hydrologic model in the relatively small Puget Sound basins, and we used the VIC model in the Yakima.

4.2.1 Implications of Climate Change on Puget Sound Catchments

We examined SWE predictions in the headwaters of the Cedar, Sultan, Tolt, and Green river basins. Figures 10 and 11 show simulated historical April 1 SWE and predicted change of SWE in the 2020s, 2040s and 2080s for A1B and B1 SRES scenarios (See Figure 9 for historical April 1 SWE). In both Figures 10 and 11, the top left illustrates the upper part of the Sultan River basin, the top right shows the upper Tolt River basin, the middle right shows the upper Cedar River basin, and the lower shows the upper Green River watershed. In the 2020s, 2040s and 2080s, the largest decrease in SWE occurs in the watershed valleys as temperature rises. Upper Cedar and Green watersheds have approximately 90% reductions in SWE in the valleys starting from the 2020s, while the Sultan and Tolt River basins, which are located in higher elevations, have smaller reductions in the 2020s. SWE decreases more substantially in the upper parts of all four basins in the 2040s, and by the 2080s, SWE is projected to disappear. Generally, simulations using the A1B SRES scenario show greater reductions in SWE (Figure 10) than those using B1 (Figure 11). Projected weekly time series of basin-averaged SWE in the four Puget Sound basins from the six composite scenarios described earlier for the 2020s, 2040s, and 2080s, as well as from all 39 ensemble scenarios for the 2020s, are summarized in Figure 12. We summarize the ensemble projections through use of a gray swath which spans the range of results from the 39 ensembles. Weekly values are summarized according to water year, October to September. The figure shows reduction of SWE throughout the winter months, compared to historical simulations. Peak SWE is projected to shift in all watersheds from near week 26 (late March), which is the average historical peak, to near week 23 (early March) by the 2020s and 2040s to near week 20 (mid-February) by the 2080s.Simulated streamflow at the reservoirs in the four basins shows a consistent shift in the hydrograph toward higher runoff in cool season and lower runoff in warm season (Figure 13). The winter peaks become higher but summer peaks become lower in the 2020s, 2040s and 2080s compared to the historical simulation. Into the future, the double-peak hydrograph

93CHAPTER 3: Hydrology and Water Resources: Washington State

Page 26: 3: Hydrology and Water Resources

Figure 9. Historical simulated April 1 snow water equivalent (SWE) in four Puget Sound watersheds (1916-2006) as simulated by the DHSVM. Watersheds are located in the overview map (smallest watershed is Tolt; Sultan, the northernmost watershed is located in the upper left corner; Green watershed is the largest; and, the Cedar watershed is in the upper right corner).

94 CHAPTER 3: Hydrology and Water Resources: Washington State

Page 27: 3: Hydrology and Water Resources

Figure 10. Projected changes in snow water equivalent (SWE) in four Puget Sound watersheds for the 2020s, 2040s, and 2080s (A1B SRES scenario) compared with simulated mean historical April 1 SWE (1916-2006) as simulated by the DHSVM. Watersheds are located in the overview map in Figure 9 (smallest watershed is Tolt; Sultan, the northernmost watershed is located in the upper left corner; Green watershed is the largest; and, the Cedar watershed is in the upper right corner).

95CHAPTER 3: Hydrology and Water Resources: Washington State

Page 28: 3: Hydrology and Water Resources

Figure 11. Projected changes in snow water equivalent (SWE) in four Puget Sound watersheds for the 2020s, 2040s, and 2080s (B1 SRES scenario) compared with simulated mean historical April 1 SWE (1916-2006) as simulated by the DHSVM. Watersheds are located in the overview map in Figure 9 (smallest watershed is Tolt; Sultan, the northernmost watershed is located in the upper left corner; Green watershed is the largest; and, the Cedar watershed is in the upper right corner).

96 CHAPTER 3: Hydrology and Water Resources: Washington State

Page 29: 3: Hydrology and Water Resources

Figure 12. Projected changes in weekly snow water equivalent (SWE) for the 2020s, 2040s, and 2080s (according to water year). Results in the top four pairs of panels are based on DHSVM simulations, while the bottom pair of panels are based on VIC model simulations. Units are meters.

97CHAPTER 3: Hydrology and Water Resources: Washington State

Page 30: 3: Hydrology and Water Resources

Figure 13. Projected changes in weekly streamflow for the 2020s, 2040s, and 2080s (A1B and B1 SRES scenarios). Results in the top four pairs of panels are based on DHSVM simulations, while the bottom pair of panels are based on VIC model simulations. Units are cubic meters per second (cms).

98 CHAPTER 3: Hydrology and Water Resources: Washington State

Page 31: 3: Hydrology and Water Resources

transforms into a single-peak hydrograph associated with increasingly rain-dominant behavior. The streamflow timing shift is mainly due to the less frequent snow occurrence, and faster and early snow melt in these historically snow-rain mixed watersheds. To assess the extent climate change might impact the timing of flow, and thus annual reservoir storage, we compared the time of year at which half of the annual (water year) flow has passed (centroid of timing, see Stewart et al, 2005). The centroid of timing (CT) values were computed from the 1917- 2006 (water year) weekly average flows. The seasonal shift is visible in the CT values (Table 8), which for the A1B emissions scenario and 2020s are about 2 weeks earlier for inflows into the Howard Hanson Reservoir on the Green River, 5 weeks earlier for Chester Morse Reservoir inflows on the Cedar River, and 3 weeks earlier for Spada Lake Reservoir on the Sultan River for the 2020s period. CT changes are smaller for B1 emissions scenarios. Given the small size (relative to mean annual inflow) of all three water supply systems, these shifts suggest that there will be increasing challenges in meeting water management objectives (Vano et al. 2009a, this report).

Puget Sound Yakima Basin

Sultan A Cedar A Tolt A Green A YAPAR

Hist 21 24 22 21 30

AIBscenarios

min 2020s 17 17 17 17 25

avg 2020s 18 19 18 19 27

max 2020s 20 21 20 20 29

2040s 17 18 17 18 24

2080s 16 16 16 17 21

B1scenarios

min 2020s 16 18 17 18 25

avg 2020s 18 20 19 19 27

max 2020s 20 22 20 20 29

2040s 17 19 18 18 26

2080s 16 17 17 17 23

* Values indicate week numbers within the water year, where:

Week 15 is Jan 7

Week 20 is Feb 11

Week 25 is Mar 18

Week 30 is Apr 22

Table 8. Centroids of streamflow timing based on weekly means for the historical period (water year 1917-2006), 2020s, 2040s and 2080s. The centroid is calculated as the time of year at which half of the annual (water year) flow has passed.

99CHAPTER 3: Hydrology and Water Resources: Washington State

Page 32: 3: Hydrology and Water Resources

4.2.2. Implications of Climate Change on the Yakima Watershed

Projections of change in April 1 SWE over the Yakima River basin are summarized in Figure 14 and indicate that for A1B and B1 emissions scenarios, respectively, SWE will decrease by 31 to 34% by the 2020s, 43 to 53% by the 2040s and 65 to 80% by the 2080s. Changes in snowpack projected for the Yakima basin are higher than projected average changes over the State as a whole (Figure 5). Weekly SWE was calculated for the Yakima watershed using results from the VIC model and are summarized in the bottom panel of Figure 12. The bottom panel of Figure 12 shows historical and projected weekly SWE for the entire Yakima River watershed. The peak weekly SWE historically occurs near week 24 (mid-March). Projections of weekly SWE for the 2020s indicate that SWE will be reduced by an average of 39 to 41% according to A1B and B1 scenarios, respectively. The peak week is projected to shift earlier to near week 23 (early to mid-March). By the 2040s, SWE will be reduced by 50 to 58% (with a peak projected to occur near week 22, or early March), and by 67 to 80% by the 2080s (with a peak projected to occur near week 20, or mid-February).We also summarized projections of weekly streamflow in the bottom panel of Figure 13 for the same suite of scenarios evaluated with respect to SWE. Peak streamflow historically occurs near week 34 (mid-May) in the Yakima River at the USGS gage at Parker. The suite of projections for the 2020s indicate that the peak streamflow will not shift significantly; however, increased streamflow in winter is expected. By the 2040s, the spring peak streamflow is projected to shift earlier near week 30 (mid- to late April) and a significant second peak flow is projected in the winter, which is characteristic of historically lower elevation transient watersheds. By the 2080s, a significant shift in the hydrologic characteristics of the watershed are projected, as the spring peak is lost and peak streamflow is projected to occur in the winter near week 20 (mid-February) which is more characteristic of rain dominant watersheds. Thus warming through the 21st century will result in increasingly rain-dominant behavior in the Yakima basin. Similar to our analysis for the Puget Sound watersheds, we evaluated the shift in the CT of flow. CT values were computed from the 1917- 2006 (water year) weekly average flows for the unregulated flow of the Yakima River at Parker, which provides a representation of naturalized flow throughout the basin. Historically, the CT occurs in mid-April (week 30). In the 2020s scenarios, the CTA seasonal shift is visible in the CT values, which for the A1B emissions scenario and 2020s is about 3 weeks earlier for both A1B and B1 scenarios. In the 2040s and 2080s for the A1B scenarios, flows shift by 6 and 9 weeks respectively. For the B1 scenarios, these shifts are 4 weeks earlier for the 2040s and 7 weeks for the 2080s. These results are summarized in Table 8. These hydrologic changes will have important implications for irrigated agriculture in WA (Vano et al., 2009b, this report).

5. Conclusions and Recommendations

Climate change will impact Washington’s hydrologic resources significantly over the next century. Sensitive areas, such as transient watersheds will experience substantial impacts by the 2020s. Annual runoff across the state is projected to increase by 0-2% by the 2020s, 2.2-2.7% by the 2040s, and

100 CHAPTER 3: Hydrology and Water Resources: Washington State

Page 33: 3: Hydrology and Water Resources

Figure 14. Projected changes in April 1 snow water equivalent (SWE) in the Yakima River basin for the 2020s, 2040s, and 2080s (A1B and B1 SRES scenarios) compared with mean historical April 1 SWE (1916-2006) as simulated by the VIC model.

101CHAPTER 3: Hydrology and Water Resources: Washington State

Page 34: 3: Hydrology and Water Resources

4.2-6.4% by the 2080s. These changes are primarily driven by projected increases in winter precipitation. April 1 SWE is projected to decrease by an average of approximately 27-29% across the state by the 2020s, 37-44% by the 2040s and 53-65% by the 2080s, based on composite changes in temperature and precipitation as summarized by Mote and Salathé (2008). Soil moisture is projected to be in the 38th to 43rd percentile by the 2020s, 35th to 40th percentile by the 2040s, and 32nd to 35th percentile by the 2080s, with 50% being equal to mean historical values. The effects of climate change on the urban water supply basins of Puget Sound and the agriculturally rich area of the Yakima basin will be significant. In the watersheds of the Puget Sound, which are characterized as transient rain-snow watersheds, snowpack is projected to decrease and seasonal streamflow is projected to shift from the characteristic double-peak to a single-peak, characteristic of rain-dominant watersheds. By the 2080s, April 1 snowpack in the watersheds will be almost entirely absent.Projections of weekly SWE over the Yakima basin indicate that it will decrease by an average of 39% by the 2020s, 50% by the 2040s, and 70% by the 2080s. The suite of projections for the 2020s indicate increased streamflow in winter but no significant change in the timing of the peak. Yet, by the 2040s, the spring peak streamflow is projected to shift toward a characteristic lower elevation transient watershed with two streamflow peaks (defined in Section 1). And by the 2080s, the streamflow regime will become rain dominant. This study utilizes climate change projections from the full suite of 39 scenarios based on A1B and B1 SRES scenarios using a delta method approach. However, further refinement of the statistical downscaling of the transient daily climate change projections such that results from coupled hydrologic simulations are robust at sub-monthly time scales would be beneficial to evaluate the potential changes in the relative variability of temperature and precipitation and other related variables. The combination of spatial and temporal statistical downscaling can introduce unrealistic storm events in the future period. One possible method to eliminate this problem is to maintain the historic sequencing of daily variability in the transient scenarios through development of a hybrid delta method and BCSD approach. These climate change projections would provide a better understanding of the uncertainty of future climate and the variability of hydrologic processes. Barriers to widespread use of climate change projections in water resources studies include the availability of data and the knowledge to effectively and appropriately use this information for specific watershed studies. The ability to educate the public about the implications of climate change is crucial, as our climate system is non-stationary and we can no longer rely on historical information alone to plan for the future.

Acknowledgments

The authors would like to thank Shraddhanand Shukla at the University of Washington for his VIC model calibration work and consultation with respect to the 1/16th degree VIC model implementation. We would also like to thank Tazebe Kiros Beyene and Elizabeth Clark at the University of

102 CHAPTER 3: Hydrology and Water Resources: Washington State

Page 35: 3: Hydrology and Water Resources

Washington for their assistance with data analysis and troubleshooting. We also thank Rob Norheim at the Climate Impacts Group for his assistance with data analysis and graphics, and the four anonymous reviewers who provided valuable input for improving this work.

References

Alberti M, Weeks R, Coe S (2004) Urban land-cover change analysis in central Puget Sound. Photogramm Eng Remote Sens 70: 1043-1052

Barnett TP, Adam JC, Lettenmaier DP (2005) Potential impacts of a warming climate on water availability in snow-dominated regions. Nature 438: 303-309

Beven KJ, Kirkby, MJ (1979) A physically based, variable contributing area model of basin hydrology. Hydrol Sci Bull 24(1): 43-69

Bonneville Power Administration, U.S. Army Corps of Engineers(USACE), NPD, U.S. Bureau of Reclamation (USBR), PNR, (1994) The Columbia River System: The inside story. Report DOE/BP-1689, published by the USACE and USBR

Bowling LC, Lettenmaier DP (2001) The effects of forest roads and harvest on catchment hydrology in a mountainous maritime environment, in land use and watersheds: human influence on hydrology and geomorphology in urban and forest areas. Wigmosta MS, Burges, SJ (ed), AGU Water Sci Appl 2:145-164

Bowling LC, Storck P, Lettenmaier DP (2000) Hydrologic effects of logging in western Washington. United States. Wat Resour Res 36: 3223-3240

Brekke LD, Miller NL, Bashford KE, Quinn NWT, Dracup JA (2004) Climate change impacts uncertainty for water resources in the San Joaquin river basin, California. J Am Water Res Assoc 40(1):149-164

Casola JH, Kay JE, Snover AK, Norheim RA, Whitely Binder LC, Climate Impacts Group (2005) Climate impacts on Washington’s hydropower, water Supply, forests, fish, and agriculture. A report prepared for King County (Washington) by the Climate Impacts Group, Center for Science in the Earth System, Joint Institute for the Study of the Atmosphere and Ocean, University of Washington, Seattle

Cherkauer KA, Lettenmaier DP (2003) Simulation of spatial variability in snow and frozen soil. Field J Geophy Res (108)D22: 8858

Christensen NS, Wood AW, Voisin N, Lettenmaier DP, Palmer RN (2004) The effects of climate change on the hydrology and water resources of the Colorado River basin. Clim Change 62:337–363

Christensen N, Lettenmaier DP (2007) A multimodel ensemble approach to assessment of climate change impacts on the hydrology and water resources of the Colorado River basin. Hydrol Earth System Sci 11: 1417-1434

Cuo L, Lettenmaier DP, Mattheussen BP, Storck P, Wiley M (2008) Hydrological prediction for urban watersheds with the Distributed Hydrology-Soil-Vegetation Model. Hydrol Proc (22)21: 4205-4213

Daly C, Neilson RP, Phillips DL (1994) A statistical-topographic model for mapping climatological precipitation over mountainous terrain. J Appl Meteorol 33: 140-158

Daly C, Gibson WP, Taylor G, Johnson GL, Pasteris P (2002) A knowledge-based approach to the statistical mapping of climate. Clim Res 22: 99-113

Dooge JC (1992) Hydrologic models and climate change. J Geophys Res (97): D3 2677-2686

Economic and Engineering Services, Inc. (2003) Watershed management plan, Yakima River basin, for Yakima River watershed planning unit and Tri-County water resources agency

103CHAPTER 3: Hydrology and Water Resources: Washington State

Page 36: 3: Hydrology and Water Resources

Hamlet AF, Lettenmaier DP (1999) Effects of climate change on hydrology and water resources of the Columbia River basin. J Am Water Resour Assoc 35:1597–1624

Hamlet AF, Mote PW, Clark M, Lettenmaier DP (2005) Effects of temperature and precipitation variability on snowpack trends in the western United States. J Clim 18(21): 4545-4561

Hamlet AF, Lettenmaier DP (2005) Production of temporally consistent gridded precipitation and temperature fields for the continental United States. J Hydrometeorol, 6: 330-336

Hamlet AF, Lettenmaier DP (2007) Effects of 20th century warming and climate variability on flood risk in the western US. Water Resour Res 43: W06427

Hamlet AF, Lee SY, Mickleson KEB, Elsner MM (2009) Effects of projected climate change on energy supply and demand in the Pacific Northwest. Washington Climate Change Impacts Assessment: Evaluating Washington’s future in a changing climate. In press

Hayhoe K, Cayan D, Field C, Frumhoff P, Maurer E, Miller N, Moser S, Schneider S, Cahill K, Cleland E, Dale L, Drapek R, Hanemann RM, Kalkstein L, Lenihan J, Lunch C, Neilson R, Sheridan S, Verville J (2004) Emissions pathways, climate change, and impacts on California. Proc Natl Acad Sci USA 101 (34):12422–12427

Hayhoe K, Wake C, Huntington TG, Luo L, Schwartz MD, Sheffield J, Wood EF, Anderson B, Bradbury J, DeGaetano TT, Wolfe D (2007) Past and future changes in climate and hydrological indicators in the U.S. Northeast. Clim Dyn 28: 381-407

IPCC (2007) Summary for policymakers. In: Climate Change 2007: The physical science basis. Contribution of working group I to the fourth assessment report of the intergovernmental panel on climate change. Solomon S, Qin D, Manning M, Chen Z, Marquis M, Averyt KB, Tignor M Miller HL (eds) Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA

Kalaney E, et al. (1996 The NCEP/NCAR 40-year reanalysis project. Bull Am Meteorol Soc 77: 437-471

Lamarche J, Lettenmaier DP (2001) Effects of forest roads on flood flows in the Deschutes River basin, Washington. Earth Surf Proc Landf 26: 115-134

Liang X, Lettenmaier DP, Wood EF, Burges SJ (1994) A simple hydrologically based model of land surface water and energy fluxes for GSMs. J Geophys Res (99)D7: 14,415-14,428

Liang X, Wood EF, Lettenmaier DP (1996) Surface soil moisture parameterization of the VIC-2L model: Evaluation and modifications. Glob Planet Chang 13: 195-206

Liang X, Wood EF, Lohmann D, Lettenmaier DP, and others (1998) The project for intercomparison of land-surface parameterization schemes (PILPS) phase-2c Red-Arkansas River basin experiment: 2. Spatial and temporal analysis of energy fluxes J Glob Planet Chang 19: 137-159

Littell JS, Oneil EE, McKenzie D, Hicke JA, Lutz J, Norheim RA, Elsner MM (2009) Forest ecosystems, disturbance, and climatic change in Washington State, USA. Washington Climate Change Impacts Assessment: Evaluating Washington’s future in a changing climate. In press

Leung L. R., Wigmosta M. S., Ghan S. J., Epstein D. J., Vail L. W. (1996) Application of a subgrid orographic precipitation/surface hydrology scheme to a mountain watershed. J Geopgys Research 101(D8): 12803-12817

Matheussen B, Kirschbaum RL, Goodman IA, O’Donnell GM, Lettenmaier DP (2000) Effects of land cover change on streamflow in the interior Columbia basin. Hydrol Proc 14(5): 867-885

Maurer EP, Wood AW, Adam JC, Lettenmaier DP, Nijssen B (2002) A long-term hydrologically based dataset of land surface fluxes and states for the conterminous United States. J Clim 15: 3237-3251

Maurer EP (2007) Uncertainty in hydrologic impacts of climate change in the Sierra

104 CHAPTER 3: Hydrology and Water Resources: Washington State

Page 37: 3: Hydrology and Water Resources

Nevada, California under two emissions scenarios. Clim Chang 82( 3-4): 309-325

Maurer EP, Duffy PB (2005) Uncertainty in projections of streamflow changes due to climate change in California. Geophys Res Let 32(3): L03704

Meehl G.A., Stocker T.F., Collins W.D., Friedlingstein P., Gaye A.T., Gregory J.M., Kitoh A., Knutti R., Murphy J.M., Noda A., Raper S.C.B., Watterson I.G., Weaver A.J., Zhao Z.-C. (2007) Global Climate Projections. In: Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change [Solomon, S., D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M. Tignor and H.L. Miller (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.

Mekis E, Hogg WD (1999) Rehabilitation and analysis of Canadian daily precipitation time series. Atmos Ocean 37: 53-85

Miller NL, Bashford KE, Strem E (2003) Potential impacts of climate change on California hydrology. J Am Water Res Assoc 39: 771-784

Milly PCD, Dunne KA, Vecchia AV (2005) Global pattern of trends in streamflow and water availability in a changing climate. Nature 438: 347-350

Mote PW, Hamlet AF, Clark M, Lettenmaier DP (2005) Declining mountain snowpack in western North America. Bull Am Meteorol Soc 86(1):39-49

Mote PW (2006) Climate-driven variability and trends in mountain snowpack in western North America. J of Clim 19(23): 6209-6220

Mote PW, Salathé Jr EP (2009) Future climate in the Pacific Northwest. Washington Climate Change Impacts Assessment: Evaluating Washington’s future in a changing climate. In press

Nakicenovic N, Swart R (eds.) (2000) Special report on emissions scenarios. A Special report of working group III of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 599

Nash LL, Gleick PH (1991) The sensitivity of streamflow in the Colorado Basin to climatic changes. J Hydrol 125: 221-241

Nijssen BN, Lettenmaier DP, Liang X, Wetzel SW, Wood EF (1997) Streamflow simulation for continental-scale river basins. Water Resour Res 33(4): 711-724

Nijssen BN, O’Donnell GM, Lettenmaier DP, Lohmann D, Wood EF (2001) Predicting the discharge of global rivers. J Clim 14: 3307-3323

Payne JT, Wood AW, Hamlet AF, Palmer RN, Lettenmaier DP (2004) Mitigating the effects of climate change on the water resources of the Columbia River basin. Clim Change 62:233–256

Refsgaard JC, Storm S (1996) Distributed hydrologic modeling. Chapter 3: Consturction, calibration, and validation of hydrological models. Kluswer Academic Publishers.

Salathé EP, Mote PW, Wiley MW (2007) Review of scenario selection and downscaling methods for the assessment of climate change impacts on hydrology in the United States Pacific Northwest. Int J Climatol 27(12): 1611-1621

Sankarasubramanian A, Vogel RM, Limbrunner JF (2001) Climate elasticity of streamflow in the United States. Water Res Res 37(6): 1771-1781

Schaake JC (1990) From climate to flow. In: Climate change and US water resources. Waggoner, PE (ed) John Wiley, New York, USA pp177-206

Shepard DS (1984) Computer mapping: The SYMAP interpolation algorithm. Spatial Statistics and Models Gaille, GL Willmott CJ (eds) Reidel, 133-145

Snover AK, Miles EL (In review) Regional integrated climate impacts assessment. In: Chapter 2, Snover, AK, Miles EL, and the Climate Impacts Group, Rhythms of change: An integrated assessment of climate impacts on the Pacific Northwest,

105CHAPTER 3: Hydrology and Water Resources: Washington State

Page 38: 3: Hydrology and Water Resources

Cambridge, Massachusetts, MIT Press

Snover AK, Hamlet AF, Lettenmaier DP (2003) Climate change scenarios for water planning studies: Pilot applications in the Pacific Northwest. Bull Am Meteorol Soc 84(11):1513-1518

Stewart IT, Cayan DR, Dettinger MD (2005) Changes toward earlier streamflow timing across western North America. J Clim 18: 1136-1155

Vano JA, Voisin N, Cuo L, Hamlet AF, Elsner MM, Palmer RN, Polebitski, A, Lettenmaier DP (2009a) Multi-model assessment of the impacts of climate change on water management in the Puget Sound region, Washington, USA. Washington Climate Change Impacts Assessment: Evaluating Washington’s future in a changing climate. In press

Vano JA, Voisin N, Scott M, Stöckle CO, Halmet AF, Mickleson KEB, Elsner MM, Lettenmaier DP (2009b) Multi-model assessment of the impacts of climate change on water management and agriculture of the Yakima River basin, Washington, USA. Washington Climate Change Impacts Assessment: Evaluating Washington’s future in a changing climate. In press

Van Rheenen NT, Wood AW, Palmer RN, Lettenmaier DP (2004) Potential implications of PCM climate change scenarios for California hydrology and water resources. Clim Chang 62:257–281

Vincent LA, Zhang X, Bonsal BR, Hogg WD (2002) Homogenization of daily temperatures over Canada. J Clim 15: 1322-1334

Vicuna S, Maurer EP, Joyce B, Dracup JA, Purkey D (2007) The sensitivity of California water resources to climate change scenarios. J Am Water Res Assoc 43(2): 482 –498

Wigmosta MS, Vail LW, Lettenmaier DP (1994) A distributed hydrology: Vegetation model for complex terrain. Can J For Res 30: 1665-1679

Wang A, Bohn TJ, Mahanama SP, Koster RD, Lettenmaier DP (2008) Multimodel ensemble reconstruction of drought over the continental United States. J Clim (accepted)

Wood AW, Leung LR, Sridhar V, Lettenmaier DP (2004) Hydrologic implications of dynamical and statistical approaches to downscaling climate model outputs. Clim Chang 62(1-3): 189-216

Wood AW, Maurer EP, Kumar A. Lettenmaier DP (2002) Long range experimental hydrologic forecasting for the eastern U.S. J Geophys Res 107(D20): 4429

Photo credit, page 71: Ron Nichols, USDA Natural Resources Conservation Service

106 CHAPTER 3: Hydrology and Water Resources: Washington State